fix: Multi byte truncate for displayer (#464)

This is needed for calling strdisplaywidth() from Lua loop.

See https://github.com/nvim-telescope/telescope.nvim/issues/414
See https://github.com/neovim/neovim/blob/a1ed941a7881122fda2fd48e71e890ed55e4d08e/src/nvim/eval/funcs.c#L9845-L9858

Co-authored-by: Simon Hauser <Simon-Hauser@outlook.de>

4 files changed, 138 insertions, 9 deletions

diff --git a/lua/telescope/make_entry.lua b/lua/telescope/make_entry.lua
index bd1bf73..d23dcb7 100644
--- a/lua/telescope/make_entry.lua
+++ b/lua/telescope/make_entry.lua
@@ -463,7 +463,8 @@ function make_entry.gen_from_buffer(opts)
 
   local icon_width = 0
   if not disable_devicons then
-    icon_width = vim.fn.strdisplaywidth(get_devicons('fname', disable_devicons))
+    local icon, _ = get_devicons('fname', disable_devicons)
+    icon_width = utils.strdisplaywidth(icon)
   end
 
   local displayer = entry_display.create {
diff --git a/lua/telescope/pickers/entry_display.lua b/lua/telescope/pickers/entry_display.lua
index c7a51ac..7e38ac8 100644
--- a/lua/telescope/pickers/entry_display.lua
+++ b/lua/telescope/pickers/entry_display.lua
@@ -1,13 +1,27 @@
+local utils = require('telescope.utils')
+
 local entry_display = {}
 
-local function truncate(str, len)
+entry_display.truncate = function(str, len)
   str = tostring(str) -- We need to make sure its an actually a string and not a number
-  -- TODO: This doesn't handle multi byte chars...
-  if vim.fn.strdisplaywidth(str) > len then
-    str = str:sub(1, len - 1)
-    str = str .. "…"
+  if utils.strdisplaywidth(str) <= len then
+    return str
+  end
+  local charlen = 0
+  local cur_len = 0
+  local result = ''
+  local len_of_dots = utils.strdisplaywidth('…')
+  while true do
+    local part = utils.strcharpart(str, charlen, 1)
+    cur_len = cur_len + utils.strdisplaywidth(part)
+    if (cur_len + len_of_dots) > len then
+      result = result .. '…'
+      break
+    end
+    result = result .. part
+    charlen = charlen + 1
   end
-  return str
+  return result
 end
 
 entry_display.create = function(configuration)
@@ -18,9 +32,9 @@ entry_display.create = function(configuration)
       local format_str = "%" .. justify .. v.width .. "s"
       table.insert(generator, function(item)
         if type(item) == 'table' then
-          return string.format(format_str, truncate(item[1], v.width)), item[2]
+          return string.format(format_str, entry_display.truncate(item[1], v.width)), item[2]
         else
-          return string.format(format_str, truncate(item, v.width))
+          return string.format(format_str, entry_display.truncate(item, v.width))
         end
       end)
     else
diff --git a/lua/telescope/utils.lua b/lua/telescope/utils.lua
index 913c9a2..9cfc23c 100644
--- a/lua/telescope/utils.lua
+++ b/lua/telescope/utils.lua
@@ -202,4 +202,86 @@ function utils.get_os_command_output(cmd)
   return Job:new({ command = command, args = cmd }):sync()
 end
 
+utils.strdisplaywidth = (function()
+  if jit then
+    local ffi = require('ffi')
+    ffi.cdef[[
+      typedef unsigned char char_u;
+      int linetabsize_col(int startcol, char_u *s);
+    ]]
+
+    return function(str, col)
+      local startcol = col or 0
+      local s = ffi.new('char[?]', #str + 1)
+      ffi.copy(s, str)
+      return ffi.C.linetabsize_col(startcol, s) - startcol
+    end
+  else
+    return function(str, col)
+      return #str - (col or 0)
+    end
+  end
+end)()
+
+utils.utf_ptr2len = (function()
+  if jit then
+    local ffi = require('ffi')
+    ffi.cdef[[
+      typedef unsigned char char_u;
+      int utf_ptr2len(const char_u *const p);
+    ]]
+
+    return function(str)
+      local c_str = ffi.new('char[?]', #str + 1)
+      ffi.copy(c_str, str)
+      return ffi.C.utf_ptr2len(c_str)
+    end
+  else
+    return function(str)
+      return str == '' and 0 or 1
+    end
+  end
+end)()
+
+function utils.strcharpart(str, nchar, charlen)
+  local nbyte = 0
+  if nchar > 0 then
+    while nchar > 0 and nbyte < #str do
+      nbyte = nbyte + utils.utf_ptr2len(str:sub(nbyte + 1))
+      nchar = nchar - 1
+    end
+  else
+    nbyte = nchar
+  end
+
+  local len = 0
+  if charlen then
+    while charlen > 0 and nbyte + len < #str do
+      local off = nbyte + len
+      if off < 0 then
+        len = len + 1
+      else
+        len = len + utils.utf_ptr2len(str:sub(off + 1))
+      end
+      charlen = charlen - 1
+    end
+  else
+    len = #str - nbyte
+  end
+
+  if nbyte < 0 then
+    len = len + nbyte
+    nbyte = 0
+  elseif nbyte > #str then
+    nbyte = #str
+  end
+  if len < 0 then
+    len = 0
+  elseif nbyte + len > #str then
+    len = #str - nbyte
+  end
+
+  return str:sub(nbyte + 1, nbyte + len)
+end
+
 return utils
diff --git a/lua/tests/automated/entry_display_spec.lua b/lua/tests/automated/entry_display_spec.lua
new file mode 100644
index 0000000..fff78cd
--- /dev/null
+++ b/lua/tests/automated/entry_display_spec.lua
@@ -0,0 +1,32 @@
+local entry_display = require('telescope.pickers.entry_display')
+
+describe('truncate', function()
+  for _, ambiwidth in ipairs{'single', 'double'} do
+    for _, case in ipairs{
+      {args = {'abcde', 6}, expected = {single = 'abcde', double = 'abcde'}},
+      {args = {'abcde', 5}, expected = {single = 'abcde', double = 'abcde'}},
+      {args = {'abcde', 4}, expected = {single = 'abc…', double = 'ab…'}},
+      {args = {'アイウエオ', 11}, expected = {single = 'アイウエオ', double = 'アイウエオ'}},
+      {args = {'アイウエオ', 10}, expected = {single = 'アイウエオ', double = 'アイウエオ'}},
+      {args = {'アイウエオ', 9}, expected = {single = 'アイウエ…', double = 'アイウ…'}},
+      {args = {'アイウエオ', 8}, expected = {single = 'アイウ…', double = 'アイウ…'}},
+      {args = {'├─┤', 7}, expected = {single = '├─┤', double = '├─┤'}},
+      {args = {'├─┤', 6}, expected = {single = '├─┤', double = '├─┤'}},
+      {args = {'├─┤', 5}, expected = {single = '├─┤', double = '├…'}},
+      {args = {'├─┤', 4}, expected = {single = '├─┤', double = '├…'}},
+      {args = {'├─┤', 3}, expected = {single = '├─┤', double = '…'}},
+      {args = {'├─┤', 2}, expected = {single = '├…', double = '…'}},
+    } do
+      local msg = ('can truncate: ambiwidth = %s, [%s, %d] -> %s'):format(ambiwidth, case.args[1], case.args[2], case.expected[ambiwidth])
+      it(msg, function()
+        local original = vim.o.ambiwidth
+        vim.o.ambiwidth = ambiwidth
+        assert.are.same(
+          case.expected[ambiwidth],
+          entry_display.truncate(case.args[1], case.args[2])
+        )
+        vim.o.ambiwidth = original
+      end)
+    end
+  end
+end)